1

XXXII Conferenza scientifica annuale AISRe

Il ruolo delle città nell’economia della conoscenza

Torino, 15-17 settembre 2011

Understanding Inappropriateness in Health Care: The Role of Supply Structure,

Pricing Policies and Political Institutions in Caesarean Deliveries ♠

Maura Francese*, Massimiliano Piacenza#, Marzia Romanelli∗ and Gilberto Turati#

June 2011 (preliminary version)

Abstract

The upward trend in the incidence of caesarean deliveries is a widespread stylised fact in many countries. Several studies have argued that it does not reflect, at least in part, patients’ needs but that it is also influenced by other factors, such as providers/physicians incentives. Not surprisingly, the incidence of caesarean sections is often used as an indicator of the degree of (in)appropriateness in health care, which has also been found to be strongly correlated with expenditure differentials between regions. We exploit the significant regional variation in the share of caesarean deliveries in Italy to explore the impact on inappropriateness of three groups of policy variables: 1) political economy indicators (as a way to capture different approaches to the governance of the health care sector); 2) reimbursement and pricing policies (as DRG fees); 3) structural supply indicators (such as the incidence of private providers and the number of employees). The analysis controls for the demographic characteristics of patients and their education levels. Results suggest that tariffs might be an effective policy tool to control inappropriateness; however, the structure of the regional health care system matters. More importantly, also some characteristics of the regional governments and the financing mechanisms play a role.

JEL Classification: H75, I18, D78, L33

Keywords: health care, inappropriateness, regional disparities, pricing policy, political economy

♠ We wish to thank B. Baltagi and A. Jones for helpful comments. M. Piacenza and G. Turati gratefully acknowledge the financial support by the Banca d’Italia. The views expressed in this paper are those of the authors and do not necessarily reflect those of the Banca d’Italia or of the University of Torino.The usual disclaimers apply. * Banca d’Italia, Structural Economic Analysis Department – Economics, Research and International Relations Area.

E-mail: maura.francese@bancaditalia.it; marzia.romanelli@bancaditalia.it. # University of Torino – School of Economics, Department of Economics and Public Finance ”G. Prato”.

E-mail: piacenza@econ.unito.it; turati@econ.unito.it.

2

1. Introduction

The expected growth in public health expenditure constitutes a relevant policy problem in almost

all developed countries. Not surprisingly, improving spending efficiency while guaranteeing (or

even improving) citizens’ health is becoming a key challenge for policy-makers. A common

suggestion to reach this goal coming from the policy-oriented literature is to improve service

appropriateness: deliver appropriate services (at the lowest possible cost) would produce cost

savings, while assuring citizens’ health. Evidence on these potential savings is provided for

instance for Italy which displays a significant variation across Italian regions in the degree of

(in)appropriateness. Moreover, this (in)appropriateness is shown to be strongly correlated with

the expenditure differentials observed between regions (Francese and Romanelli, 2010).

Improving the appropriateness of medical treatments could then clearly contribute to the efforts

of containing public health expenditure, without reducing or limiting the quantity or quality of

services to be supplied to patients.

The incidence of cesarean sections is an indicator of (in)appropriateness commonly

considered in the literature, and by policy-makers1. Being a surgical treatment, a caesarean

section is characterised by a large cost differential with respect to the alternative classical vaginal

delivery (a medical treatment). Absent any therapeutic reasons, this latter treatment is generally

considered an appropriate way of delivery, which can clearly help in containing health care costs.

An upward trend in the incidence of caesarean deliveries is a well documented stylised fact

at the international level. The main explanations proposed by the literature focus on the role of

many different factors: from technological changes (affecting for instance the treatment of pain in

delivery), to changes in patients’ preferences and the physicians/providers behaviours (the latter

being also influenced by the remuneration system).

A remarkable rising trend in the rate of caesarean sections has also characterised Italy,

where this indicator has more than trebled from 1980 to 2007. Not surprisingly, the need to

monitor its dynamics has drawn the attention of policymakers. For example, the 2003-05

National Health Plan stated the objective of containing the average share of caesarean deliveries

in Italy at about 20% by the end of the planning period.2 Over those years, however, the

increasing trend in the incidence of caesarean births did not stop, bringing the Italian average to

exceed 38% in 2005. Regional variation (both in the growth rates and the incidence rates) is also

1 See for example the indicators regularly published by the Italian Health Ministry (Ministero della Salute, Rapporto annuale sull’attività di ricovero ospedaliero) and Fortino et al. (2002). 2 In particular the plan included among its objectives the aim to decrease the frequency of caesarean deliveries and reduce regional diffentials (p. 82). The stated goal was to achieve – by the end of the three years period – a national average equal to 20%, in line with mean values for other European countries. The reduction was to be obtained also through a revision of DRG reimbursment fees.

3

significant: for instance, caesarean section rates are above 50% and 60% in Sicilia and Campania

respectively, two regions characterised by high per capita public health expenditure and where

the room for cost savings is estimated to be large (Piacenza and Turati, 2010 and Francese and

Romanelli, 2010).

This paper addresses the issues of what affects (in)appropriatness (proxied by caesarean

section rates). Besides more traditional variables, we analyse also the role of the characteristics of

the Regional governments. As we will discuss below, given a national regulatory framework,

health policies are implemented and managed by Regions in a way that reflects a complex net of

intergovernmental relationships between the Central and the Regional governments. The modern

fiscal federalism theory suggests that the way in which different layers of government interacts

affect policy outcomes. Exploiting available data on caesarean deliveries, we disentangle the

impact of three groups of policy variables: 1) political economy variables, in order to capture

different approaches to the governance of the health care sector; 2) reimbursement and pricing

policies, such as DRG fees to control for the effect of financing mechanisms; 3) structural supply

indicators, to take into account the impact of different organisational arrangements. We also

provide checks for the presence of imitative behaviour driven by geographical proximity, and

serial correlation. Our results suggest that the “quality” of Regional governments and structural

characteristics of the health sector do matter.

The remainder of the paper is structured as follows. Section 2 provides a brief survey of the

available literature on caesarean sections. Section 3 presents the Italian case. The model and the

empirical strategy are discussed in Section 4. We describe the data and our results in Section 5

and 6 respectively. A brief section of concluding remarks follows.

2. Why are caesarean sections on the rise? A brief survey

The incidence of caesarean deliveries is characterised by an upward trend at the international

level. For instance among the OECD countries, in the US the caesarean section rate has increased

by about 9 percentage points over the period 1990-2006; the increase in Germany and Spain has

been respectively about 13 and 12 percentage points (OECD Health Data, 2010). Given this

evolution over time, it is not surprising that the impact of caesarean sections on maternal and

perinatal health has drawn the attention of scholars, international organisations, and national

policymakers (see for example Lumbiganon et al., 2010 on the 2007-08 WHO global survey).

The economic literature has been mainly concentrated on identifying the drivers of the observed

upward trend. Many factors help explaining the increase in the incidence of caesarean deliveries

(Ecker and Frigoletto, 2007). They can be summarised under three main categories:

4

1. technological changes: such as the move from home to hospital delivery, the use of

anaesthesia (and improved anaesthetic techniques), the introduction of modern antibiotics, the

creation of blood banks, neonatal intensive care units, techniques for monitoring the fetus

health during pregnancy and labour, and for inducing labour;

2. changes in patients’ preferences: nowadays patients might be willing to accept a lower risk of

an adverse outcome to avoid a caesarean delivery. The way in which they balance risks and

assess risk levels associated with the different procedures have changed. This might reflect

both social and cultural factors as well as changes in reproductive behaviour. For example,

the age of the mother has significantly increased with respect to the past and parents

educational levels and employment statuses (particularly for mothers) have experienced

dramatic changes3. Furthermore fertility rates and household’s size and composition are now

significantly different than just a few decades ago.

3. changes in physicians/providers behaviour: organisational characteristics of the health sector

and medical best practices have displayed large changes over the last decades. Such changes

are often linked to technological progress. However, many times they are also driven by

physicians/providers behaviours induced by exogenous factors. An obvious example are the

incentive effects of payment systems. In this category one may also consider the more intense

use of induced labours, which might reflect the scheduling of deliveries to suit providers

timetables (such as physicians and obstetricians work shifts).4 Another issue that has received

attention by the literature is the increasing fear of malpractice lawsuits, which might have

influenced physicians decisions (see for example Localio et al., 1993 and Dubay et al., 1999).

Several studies have analysed one or more of the drivers accountable for the increase in

caesarean section rates. Gruber and Owings (1996) for example investigate the impact of

declining fertility trends on the rapid rise of caesarean births. They study how much the

exogenous income shock induced on physicians incomes by the drop in fertility might have

influenced doctors’ decisions, inducing the substitution of a cheaper practice with one

characterised by a higher reimbursement. Such a reaction would be consistent with a model of

induced demand (see, e.g., the model developed by McGuire and Pauly, 1991). The authors find a

3 For example in Italy the average age of the mother has increased from 27.5 in 1980 to 31.6 in 2008, the share of women with tertiary education has almost trebled (from 4.9 per cent in 1993 to 12.8 in 2007) and female labour force participation has increase form 41.9 per cent in 1993 to 51.6 in 2008. 4 Brown (1996) examines the impact of physicians demand for leasure on caesarean section rates, observing that the probability of a caesarean delivery over the weekend and at certain hours of the day (and night) is significantly different (lower) than that for unplanned births.

5

positive, even though small5, relation between the fall in fertility and the incidence of caesarean

births.

In general, the role of fee differentials has drawn most of the economists’ attention. For

instance, Gruber, Kim and Mayzlin (1999) find that fee differentials have a positive effect on the

probability of caesarean delivery for Medicaid enrolees; furthermore they estimate that the larger

differentials for patients that are privately insured with respect to Medicaid enrolees accounts for

between ½ and ¾ of the differential in the rate of caesarean births in the two groups. Grant

(2009), which replicates Gruber, Kim and Mayzlin (1999) analysis, estimates an effect which is

about ¼ of that reported originally, the difference being mainly due to the sample and estimation

procedure adopted. According to Grant’s analysis, other factors account for most of the

difference observed between the two populations, in particular risk factors and non-random

matching between privately insured patients and providers which are more inclined to resort to

caesarean deliveries (Grant, 2005).

The issue of assortative matching between mothers and providers is also explored by Fabbri

and Monfardini (2008) on a sample of Italian patients. In particular, the authors consider two

types of providers: public and private hospitals, with the latter being characterised by a higher

inclination to resort to a caesarean section.6 According to the authors’ findings, the assortative

matching between patients and providers is of minor relevance, while the selection mechanism of

patients into hospitals is largely driven by risk factors (with the more risky patients being

admitted into public – higher quality – hospitals7). Furthermore after controlling for selection and

other observable characteristics, the probability of resorting to a caesarean delivery is higher in

private hospitals than in public ones. A finding that leaves open the issue of measuring the impact

of financial incentives which is not estimated in this study. The significant difference in the

reimbursement fee between vaginal and caesarean delivery in Italy is discussed in Pizzo (2008).

The author does not estimate the impact of fee differentials on the probability of observing a

caesarean birth. However, after noting the wide regional variation in caesarean delivery rates,

Pizzo attempts to estimate the potential savings that would been observed had the caesarean

section incidence been equal to a given reference value8.

5 About 1/6 of the total change in the caesaren section rate. 6 It should be noted that the Italian National Health System includes both public and private hospitals (the latter being subject to a formal procedure for being recognised within the system). At the national level, the share of beds in private hospitals is about 16 per cent, with significant variations across Regions. Each patient is free to choose the hospital which will provide the treatment (with no difference in out-of-pocket expenditure); the hospital will then be directly reimbursed by the competent level of government. 7 It should be noted that often in Italy private hospitals do not have emergency surgical capacity or intensive care units. 8 For example the reference value proposed by the WHO or that computed by isolating deliveries which are not characterised by risk factors commonly associated with the need to proceed with a caesarean section.

6

An evaluation of the relevance of financial incentives, supply structure indicators, and the

“quality” of governments in charge of managing health care policies in explaining the rapid

increase in caesarean delivery rates in Italy and its wide regional variation is still missing in the

literature. This work is an attempt to contribute in filling this gap.

3. Setting the stage: preliminary evidence on caesarean deliveries in Italy

With respect to trends registered at the international level, the increase in the caesarean delivery

rate in Italy has been remarkable (fig. 1). The rate in 2007 was almost 3.5 times the value

observed in 1980. Its dynamics showed significant regional variation (fig. 2, panel a, b, c) and a

clear geographical pattern: Southern regions - which in 1980 displayed values below the national

average - showed a much faster increase in the caesarean section rate, which reached maximum

values above 50% and 60% in Sicilia and Campania respectively.

A first glance at the data shows that the more frequent use of caesarean sections was

accompanied by changes in patients characteristics. The correlation between the incidence of

caesarean births and the number of patients with complications at the moment of delivery is

positive (fig. 3). The same is true if one considers the mother’s age (fig. 4), which rose

considerably over the last decades. However, the impact of changes in patients’ characteristics

have been different in the different areas of the country. For example, a simple OLS regression of

the incidence of caesarean deliveries on the mother’s age shows a statistically significant

difference in the coefficient between the South (for which the estimated coefficient is higher) and

the rest of the country, while differences between the regions in the North and in the Centre are

not statistically significant (fig. 5). This raises the questions of what other institutional or policy

factors can account, at least in part, for such evidence.

However, to understand which factors can play a role, one should have in mind the

institutional characteristics of the Italian NHS. This is a public universalistic scheme – founded in

1978 – to provide all citizens a set of compulsory health care services, involving a complex net of

intergovernmental relationships between different layers of governments (see, e.g., France et al.,

2005). In particular, while funding of the NHS is mostly in the hand of the Central government

(despite some recent moves towards a higher degree of fiscal decentralisation), the management

of the services is devolved at the regional level. Management of the services includes for instance

decisions on the network of hospitals and their staffing, the purchase of some services from

private providers, the definition of regional tariffs within the nationally defined reimbursement

mechanism (that, starting from 1995-1997, is a prospective payment system based on DRG).

Regional variability in this factors can help explain the differential trends observed in our data.

For instance, a number of studies (e.g., Fabbri and Monfardini, 2008) pointed out a higher

7

propensity for caesarean sections in private hospitals. We as well do find in our data a positive

correlation between the incidence of caesarean deliveries and the relevance of private hospitals

(measured as the share of bed in private hospitals on the total number of hospital beds – fig. 6).

From a preliminary look at the data, however, it is not easy to highlight the impact of other policy

variables. This is the case for example of reimbursement policies. If one separates the regions that

have established their own DRG tariffs from those whose reimbursement levels are in line with

national DRG tariffs, a clear pattern is not immediately traceable (fig. 7). Moreover, other

variables such as the “quality” of regional governments can affect the incidence of caesarean

sections. In order to disentangle the role of these many different factors, we therefore proceed to

the estimation of a reduced form model whose specification is outlined in the next section.

4. The model

We estimate the following model on the sample of Italian regions9 over the years (1996-2006):

(1) ∑∑∑∑∑=====

++++++++=H

hit

hit

zh

K

k

kit

kF

f

fit

wf

J

j

jit

xj

T

tt

tiit zkwxdy

11111

εβββββαα

The dependent variable ( ity ) is the log of the odd ratio of the share of caesarean deliveries in

region i in year t; iα and dt are respectively regional and year dummies. Regressors are grouped

into the following categories:

a) j = 1, …, J control variables ( jitx ) such as demographic characteristics of patients and their

education levels. These variables should proxy the ‘demand’ for caesareans by capturing the

effect of patients characteristics, health status and preferences;

b) f = 1, …, F structural supply indicators ( fitw ) such as the incidence of private providers,

hospital characteristics and the composition of the workforce, in order to control for

organisational and structural differences which could influence physicians choices;

c) k = 1, …, K reimbursement and pricing policies indicators: for this category we consider

variables related to the DRG fees ( kitk );

d) h = 1, …, H political economy indicators ( hitz ) as a way to capture the influence of regional

government characteristics and their quality.

9 We consider all the Italian regions (ordinary and special statute) and the two autonomous provinced of Trento and Bolzano. Overall we therefore have 21 regions, 4 in the North West (Piemonte, Valle d’Aosta, Lombardia, Liguria), 5 in the North East (Bolzano, Trento, Veneto, Friuli Venezia Giulia, Emilia Romagna), 4 in the Center (Toscana, Umbria, Marche, Lazio), 6 in the South (Abruzzo, Molise, Campania, Puglia, Basilicata, Calabria) and 2 Islands (Sicilia, Sardegna).

8

The analysis also checks for the presence of imitative behaviours among regions driven by

geographical proximity, and take into account the potential presence of serial correlation in the

data. Equation (1) is estimated using a panel fixed-effect estimator.10

5. The data

Economic and socio-demographics data and health status variables are drawn form various Istat

publications, while information on structural characteristics of the hospital sector are published

by the Italian Health Ministry.

As for the DRG tariffs variables, we have used the data on national tariffs as established by

the decree DM 30/06/1997 and for the more recent years as published on the Agenas11 website.

To identify the years in which each regional government has introduced regional DRG tariffs (if

they have decided to do so) we have followed the reconstruction presented in Carbone et. al.

(2006).12

The data on the characteristics of the regional governments are derived from the archives

on regional and local administrations published on-line by the Italian Ministry for Domestic

Affairs. Further information has been requested directly to regional administrative offices.

Descriptive statistics for the variables used in the empirical analysis for the years 1996-

2006 are reported in Tab. 1. Fig. 8 sketches the period over which each region has adopted their

own DRG tariffs.

6. Empirical analysis

The empirical analysis is presented by addressing first the methodological issues, in particular the

possible presence of serial and spatial correlation. The section that follows discusses our main

results.

6.1. Methodological issues

As a baseline approach we estimate equation (1) using a panel fixed-effect estimator. Tab. 2

reports the results.

We tested the adopted specification and estimation strategy in several ways. In particular

we controlled for two kinds of problems: serial and spatial correlation.

10 We performed a Hausman test comparing fixed and random-effect estimators; the results suggest the use of the former. 11 Agenzia Nazionale per i Servizi Sanitari Regionali (http://www.agenas.it/). 12 We have also requested DRG tariffs time series directly to each regional administration, which are the moment not yet available. Updates of this work will take into account any progress in data collection from regional administrations.

9

Concerning serial correlation, we performed three tests. While the Woolridge (2002)

statistics does not reject the hypothesis of no serial correlation, the Bargava et al. (1982) modified

DW and the Baltagi and Wu (1999) LBI statistics suggest the presence of serial correlation.

Therefore for all estimated specifications robust standard errors are provided. Furthermore we

extended the model to include time lagged regressors, to check for the presence of a dynamic

dimension of the model which could be due to time persistence in agents behaviour. None of

them turned out to be significant.

As for the problem of spatial correlation, we recognize that in a decentralised setting – such

as that of regional governments in Italy – the economic policies of neighbouring jurisdictions

may show a certain degree of correlation, as highlighted in several empirical studies (e.g.,

Brueckner, 2003). This interaction can be the result of political opportunism. The intuition behind

this motivation is related to the presence of private information about either the quality of the

incumbent or the costs and benefits of the policies implemented. Citizens can get some

information by comparing the performance of their politicians with the performance of politicians

in neighbouring jurisdictions; as a consequence, the incumbent government would mimic

neighbouring jurisdictions’ policies – e.g., health care policies – in order not to lose political

consensus. This strategic behaviour was firstly described by Salmon (1987) in terms of yardstick

competition and has been investigated empirically by several authors (e.g., Elhorst and Fréret,

2009; Foucault et al., 2008). In order to control for these effects, we estimated a spatial lag and a

spatial error model considering the same regressors as in column E of Tab. 1, including regional

fixed effects. The weighting matrix was computed on the basis of the Euclidean distances

between the capitals of the regions13; we considered both a row standardised and not standardised

version (Anselin, 1988). When using the latter weighting matrix the hypothesis of spatial

correlation is always rejected, while when considering the row standardised version results are

mixed, given that the hypothesis of spatial correlation is not rejected. However the magnitude,

sign and significance of the coefficients is generally confirmed14.

All in all our baseline approach seems adequate. We therefore proceed with the discussions

of the main results.

13 Distances have been calculated using the google maps distance calculator; this tool measures distances ”as the crow flies”. 14 Results for the serial correlation tests and for the spatial error and spatial lag models are available upon request.

10

6.2. Estimation results

Tab. 2 compares estimation results when adding the regressors categories listed above. We start

by considering control variables only. As expected the coefficient of the mother age is positive

while the birth rate coefficient is negative and significant. This is in line with the findings in

Gruber and Owings (1996) where a drop in the fertility rate is accompanied by an increase in

caesarean sections. The rationale is that a drop in the hospital revenue (due to the reduced number

of births) triggers a substitution between vaginal births (reimbursed at a cheaper rate) and

caesarean births (which are accompanied by a higher tariff). Another possible explanation for the

negative impact of the birth rate on caesarean deliveries may be the presence of a learning effect,

i.e., the greater experience gained by hospitals when the number of births increases which should

imply a reduced need for caesarean sections. As in Gruber and Owings (1996), we also take into

account a measure of the underlying riskiness of births, in particular by including as regressor the

neonatal mortality rate within 6 days15, which could be a factor of a more intense use of caesarean

sections. As concerns supply indicators, the incidence of medical staff on the total number of

employees positively affects the number of caesareans, while the share of beds in private

hospitals is not statistically significant16.

The presence of regional DRG reimbursement levels might be a signal that the region is

putting effort in managing health expenditure using the policy tools at its disposal. Indeed, the

dummy variable accounting for regional tariffs has a negative impact on the number of caesarean

deliveries17. However when interacting the dummy for the presence of regional tariffs with the

relative size of the private hospital sector the positive effect of regulating reimbursements is

mitigated (or even reversed)18. This suggests a note of caution: deviating from the national

reimbursement levels does not per se imply superior outcomes. When the share of private

providers is very large, it could reflect lobbying efforts and result in a worsening of the incentive

mechanisms. This interpretation is reinforced by the evidence on the effects of introducing the

tariffs.

15 The neonatal mortality rates is defined as the fraction of live births that die within 6 days. The results are confirmed also using the neonatal mortality rate at 29 days. 16 We also controlled for a measure of use intensity of hospitals facilities (average stay in hospital). The variable is not significant and does not affect neither the magnitude nor the signficance of the other coefficients. Similarly, including the ratio of beds on population (as a measure of productive capacity) yealds a non significant coefficient and does not alter the other findings. 17 This dummy variable is essentially a Centre-North dummy, in the South only one region has deviated from the level set for national tariffs. 18 The estimates for specification E in Tab. 1 imply that when the share of beds in the private hospitals is larger than 20%, the introduction of regional tariffs does, other things being equal, result in an increase in the odd ratio of caesarean deliveries.

11

Indeed the positive and significant sign associated with the variable accounting for the

introduction of regional tariffs19 suggests that the ability of keeping under control the number of

caesarean deliveries requires some time to become thoroughly effective. In the first year there are

some ‘adjustment costs’ which mitigate the positive effect of regulating the reimbursement of

medical treatments. However, the larger is the share of private providers the lower are the

adjustment costs suggesting that a wider private sector might push for a change in reimbursement

levels as an effect of lobbying efforts. Unfortunately given that we lack complete series for

reimbursement fees for the different types of treatments (caesarean and vaginal delivery) for all

the regions, we are not able at this stage to isolate the impact on the incidence of caesareans of a

one euro increase or decrease in the payment.

If we consider the characteristics of the regional government and of the president of the

region we see that the president’s experience (measured as the number of years the president has

been in charge) seems relevant. In particular a more experienced president helps containing the

inappropriateness of treatments. This results is in line with recent findings of the literature on

electoral discipline of the duration of legislative terms, which show that longer terms tend to

improve governments’ performance (e.g., Dal Bó and Rossi, 2008), mainly due to the possibility

for the legislator to devote more resources for facing relevant policy issues (e.g., increasing the

appropriateness of health care treatments). The political alignment with the central government

matters as well: the interaction between the president experience and the dummy variable

capturing political alignment between the local and central government is positive and

significant, suggesting a loosening of the pressure to control inefficiencies. This finding can be

interpreted in terms of increased president’s expectations of a more ‘benevolent’ treatment in

terms of deficit bailout by a friendly central government than by an adversary one.20

The share of own regional funding on the total resources spent for health displays a

negative sign and it is significant. This is coherent with several interpretations (and their mix).

First, in line with modern theory of fiscal federalism (e.g., Weingast, 2009), a higher degree of

fiscal autonomy determines higher electoral accountability, leading to a tighter control of

government spending and increased efficiency. From this perspective, the result adds to recent

empirical literature investigating how decentralization and vertical fiscal imbalance affect

government size (e.g., Jin and Zou, 2002; Fiva, 2006; Borge and Rattsø, 2008). Second, the

regional differences in the share of own funding mostly reflect the tax base distribution and GDP

19 The variable is equal to 1 when regional tariffs are introduced and it is equal to zero otherwise. 20 See Arulampalan et al. (2009) for further discussion on this issue.

12

inequality across the country; the variable might then capture north/south differences (de Matteis

and Messina, 2010).

In general, the signs of the coefficients and their significance levels are quite robust across

specifications. Finally, it must be noted that the empirical analysis point to the presence of

regional fixed effects of significant magnitude. The geographical pattern is as expected, with

Northern regions displaying, ceteris paribus, a lower odds ratio.

7. Concluding remarks

The goal of expenditure containment can be achieved by reducing inefficiencies through an

increase of the appropriateness of treatments. Changes in the structure and level of fees paid to

providers can contribute. Attention must be paid to providers behavioural responses, but also to

the impact on care quality and health outcomes which are not discussed in this work.

This work considers the case of caesarean deliveries in Italy. The study analyses the effects

of the presence of regional DRG tariffs as an effective policy tool to control inappropriateness.

The results suggest that the effectiveness of differentiating the reimbursement mechanism from

the national setting does not guarantees superior outcomes. The structure of the regional health

care system (for example the relevance of the private sector) does affect the outcome. The

experience and the stability of regional administrators can also play a role. Furthermore having

access to significant own resources for financing health expenditure seems to provide the right

incentives to regional governments.

13

References

Anselin, L. (1988), Spatial Econometrics: Methods and Models, Kluwer Academic Publishers,

Boston.

Arulampalan W., Dasgupta S., Dhillon A. and Dutta B. (2009), “Electoral goals and centre-state

transfers: a theoretical model and empirical evidence from India”, Journal of Development

Economics, 88, 103–119.

Baltagi, B.H. and P.X. Wu (1999), “Unequally Spaced Panel Data Regressions with AR(1)

Disturbances”, Econometric Theory, 15, 814-823.

Bhargava, A., L. Franzini and W. Narendranathan (1982), “Serial Correlation and the Fixed

Effects Models”, Review of Economic Studies, 49, 1982, 533-549.

Borge, L.E. and Rattsø, J. (2008), “Property taxation as incentive for cost control: Empirical

evidence for utility services in Norway”, European Economic Review, 52(6), 1035-1054

Brown III, H.S. (1996), “Physician Demand for Leisure: Implications for Caesarean Section

Rates”, Journal of Health Economics, 15, 233-242.

Brueckner, J.K. (2003), “Strategic Interaction among Governments: An Overview of Empirical

Studies”, International Regional Science Review, 26, 175-188.

Carbone, C., C. Jommi and A. Torbica (2006), “Tariffe e finanziamento dell’innovazione

tecnologica: analisi generale e focus su due casi regionali”, in Anessi Pessina, E. and E.

Cantù, Rapporto OASI 2006 - L’aziendalizzazione della Sanità in Italia, EGEA, Milano.

Dal Bó, E. and M. Rossi (2008), Term Length and Political Performance, NBER Working Paper,

No. 14511.

De Matteis, P. and G. Messina (2010), “Le capacità fiscali delle regioni italiane”, Rivista

eEconomica del Mezzogiorno, 3.

Dubay, L., R. Kaestner and T. Waidmann (1999), “The Impact of Malpractice Fears on Cesarean

Section Rates”, Journal of Health Economics, 18, 491-522.

Ecker, J.L. and F.D. Frigoletto (2007), “Cesarean Delivery and the Risk-Benefit Calculus”, The

New England Journal of Medicine, 356 (9), 885-888.

Elhorst, J.P. and S. Fréret (2009), “Evidence of Political Yardstick Competition in France Using a

Two-regime Spatial Durbin Model with Fixed Effects”, Journal of Regional Science, 20(10),

1-21.

14

Fabbri, D. and C. Monfardini (2008), “Style of Practice and Assortative Mating:a Recursive

Probit Analysis of Caesarean Section Scheduling in Italy”, Applied Economics, 40, 1411-

1423.

Fiva, J. (2006), “New Evidence on the Effect of Fiscal Decentralization on the Size and

Composition of Government Spending”, FinanzArchiv – Public Finance Analysis, 62(2), 250-

280.

Fortino, A., L. Lispi, F. D’Ippolito and G. Ascone (2002), L’Eccessivo L’eccessivo Ricorso

ricorso al Taglio taglio Cesareo cesareo – Analisi dei Dati dati Italianiitaliani, Ministero

della Salute, available at

http://www.salute.gov.it/ricoveriOspedalieri/archivioDocumentiRicoveriOspedalieri.jsp?lingu

a=italiano&menu=documenti.

Foucault, M., T. Madies and S. Paty (2008), “Public Spending Interactions and Local Politics.

Empirical Evidence from French Municipalities”, Public Choice, 137, 57-80.

France, G., Taroni, F. and A. Donatini (2005), “The Italian Health-care System”, Health

Economics, 14, 187–202.

Francese, M. and M. Romanelli (2010), “Health Care in Italy: Expenditure Determinants and

Regional Differentials”, in A. Testi et al., Proceedings of the XXXVI International ORHAS

Confernce – Operations Research for Patient – Centered Health Care Delivery, Franco

Angeli, Milano; and paper presented at the 3rd Biennial Conference of the American Society

of Health Economists, Cornell, Ithaca (NY).

Grant, D. (2005), “Explaining Source of Payment Differnces in U.S. Cesarean Rates: Why Do

Privately Insured Mothers Receive More Cesareans than Mothers Who Are Not Privately

Insured?”, Health Care Management Science, 8, 5-17.

Grant, D. (2009), “Physician Financial Incentives and Cesarean Delivery: New Conclusions from

the Healthcare Cost and Utilization Project”, Journal of Health Economics, 28, 244-250.

Gruber, J., J. Kim J. and D. Mayzil (1999), “Physician Fees and Procedure Intensity: the Case of

Cesarean Delivery”, Journal of Health Economics, 18, 473-490.

Gruber. J. and M. Owings (1996), “Physician Financial Incentives and Cesarean Section

Delivery”, Rand Journal of Economics, 27, 99-123.

Jin, J. and Zou, H. (2002), “How does fiscal decentralization affect aggregate, national, and

subnational government size?, Journal of Urban Economics, 52, 270-293.

15

Localio, A.R., A.G. Lawthers, J.M. Bengtson, L.E. Hebert, S.L. Weaver, T.A. Brennan and J. R.

Landis (1993), “Relationship Between Malpractice Claims and Cesarean Delivery”, The

Journal of the American Medical Association, 269(3), 366-373.

Lumbiganon, P., M. Laopaiboon, A. M. Gulmezoglu, J. P. Souza, S. Taneepanichskul, P. Ruyan,

D. E. Attygalle, N. Shrestha, R. Mori, N. D. Hinh, H. T. Bang, T. Rathavy, K. Chuyun, K.

Cheang, M. Festin, V. udomprasertgulUdomprasertgul, M. J. V. Germar, G. Yanqiu, M, Roy,

G. Carroli, K. Ba-Thike, E. Filatova, and J. Villar (2010), “Method of Delivery and

Pregnancy Outcomes in Asia: the WHO Global Survey on Maternal and Perinatal Health

2007-08”, Lancet, 375, pp. 490-499.

McGuire, T. and M.V. Pauly (1991), “Physician Response to Fee Changes with Multiple Payers”,

Journal of Health Economics, 10, 385-410.

Ministero della Salute (2003), Piano Sanitario Nazionale 2003-2005, Roma.

Ministero della Salute (various years), Rapporto annuale sull’attività di ricovero ospedaliero,

Roma.

OECD (2010), Health data, Paris.

Piacenza, M. and G. Turati (2010), Does fiscal discipline towards sub-national governments

affect citizens’ well-being? Evidence on health, Working Paper n. 12, Department of

Economics and Public Finance G. Prato, University of Torino.

Pizzo, E. (2008), “Sostenibilità dei sistemi sanitari pubblici ed appropriatezza delle cure: la

pratica medica dei parti cesarei in Italia e nel Regno Unito”, Proceedings of the International

Conference “Social Sciences and Health in the 21st Century: New Trend Old Dilemmas?”,

Milano, Franco Angeli, 285-303.

Salmon, P. (1987), “Decentralisation as an Incentive Scheme”, Oxford Review of Economic

Policy, 3(2), 24–43.

Weingast, B.R. (2009), “Second Generation Fiscal Federalism: the implication of fiscal

incentives”, Journal of Urban Economics, 65, 279-293.

Wooldridge, J. (2002), Econometric Analysis of Cross Section and Panel Data, MIT Press.

16

Table 1

# obs mean std. dev. min max

dependent variable odds ratio of caesarean deliveries 210 -0.740 0.435 -1.805 0.426mother's age 210 30.799 0.933 28.500 33.880birth rate 189 9.167 1.205 6.900 12.230% primary school educ (females) 210 17.715 3.700 9.680 26.244neonatal mortality (first 6 days) 210 11.108 5.300 0.000 37.105medical staff (per 1000 residents) 210 54.753 3.361 45.671 61.607bed in private hospitals (ratio) 189 11.754 8.381 0.000 35.051share of own funding 210 0.376 0.149 0.065 0.728president experience 210 3.314 3.157 0 16

control variables (x)

structural supply indicators (w)political economy indicators (z)

Descriptive statistics

17

Table 2

18

Figures

Figure 1 – Caesarean deliveries (%)

Cesarean deliveries (%)

0

10

20

30

40

50

60

1980

1982

1984

1986

1988

1990

1992

1994

1996

1998

2000

2002

2004

2006

Note: continous lines North; broken lines Center; lines with circles South& Islands

Piemonte

Valled'Aosta

Lombardia

P.A.Bolzano

P.A.Trento

Veneto

Friuli-VeneziaGiulia

Liguria

Emilia-Romagna

Toscana

Umbria

Marche

Lazio

Abruzzo

Molise

Campania

Puglia

Basilicata

Calabria

Sicilia

Sardegna

Italia

9

19

Figure 2

a) North

Cesarean deliveries - North(normalised with respect to national average)

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

1980

1982

1984

1986

1988

1990

1992

1994

1996

1998

2000

2002

2004

2006

Note: continous lines North; broken lines Center; lines with circles South& Islands

Piemonte

Valled'Aosta

Lombardia

P.A.Bolzano

P.A.Trento

Veneto

Friuli-VeneziaGiulia

Liguria

Emilia-Romagna

Italia

b) Centre

Cesarean deliveries - Centre(normalised with respect to national average)

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

1980

1982

1984

1986

1988

1990

1992

1994

1996

1998

2000

2002

2004

2006

Note: continous lines North; broken lines Center; lines with circles South& Islands

Toscana

Umbria

Marche

Lazio

Italia

c) South & Islands

Cesarean deliveries - South and Islands(normalised with respect to national average)

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

1980

1982

1984

1986

1988

1990

1992

1994

1996

1998

2000

2002

2004

2006

Note: continous lines North; broken lines Center; lines with circles South& Islands

Abruzzo

Molise

Campania

Puglia

Basilicata

Calabria

Sicilia

Sardegna

Italia

20

Figure 3

% of caesarean and discharges with complications

020

4060

ces

5 6 7 8dim_com_parto

Figure 4

% of cesarean and average mother’s age

020

4060

ces

26 28 30 32 34mother_age

Figure 5

% of caesarean and mother age by macro area

020

4060

26 28 30 32 34mother_age

ces Fitted valuesces Fitted valuesces Fitted values

Nord=green Centre=red South&Islands=blue

21

Figure 6

% of caesarean and share of beds in private hospitals

020

4060

ces

0 10 20 30 40share_plo_priaccr

Figure 7

% of caesarean & average mother age

in region which have established their own DRG tariffs (green)

and in regions aligned with national DRG tariffs (red)

1020

3040

5060

ces

28 30 32 34mother_age

ces ces

squares - green: Regions with regional DRG tariffs and circles - cranberry: Regions aligned to national DRG tariffs

22

Figure 8

Time line of the introduction of regional tariffs in the period 1997-2009

(a red line means the region has its own DRG tariffs)

Piemonte Valle d’Aosta

Lombardia

Bolzano Trento Veneto

Friuli VG Liguria

Emilia Romangna

Toscana

Umbria

Marche

Lazio

Abruzzo

Molise

Campania

Puglia

Basilicata

Calabria

SiciliaSardegna

1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009